Why Bookmarks Break: The Six Ways a Saved Link Dies

Here is what nobody mentions when you press Ctrl+D: you have not saved anything. You have written down an address. Whether there is still a building at that address next year is not your decision, and you will usually find out at the exact moment you needed it.

Link rot is the general name for this, and it is not one problem. It is six, with different causes and different defences. Knowing which one you are looking at tells you whether anything can be done.

1. The author deleted it

The most common, and the least preventable. Someone posts something useful, thinks better of it a month later, and removes it. On social platforms this happens constantly — a candid thread about a layoff, a frank post about pricing, a take that aged badly.

There is a particularly awkward variant: the author deletes their whole account, and every post goes at once. On Reddit the body can vanish while the title stays, leaving something that looks alive from the outside and is empty when you open it.

2. The URL changed

Sites restructure. A blog moves from /2019/03/post-title to /blog/post-title. A product page moves under a new category. A company rebrands and takes the whole path structure with it. Careful sites redirect; most do not, or they redirect for a year and then stop renewing the rule.

3. The site is gone

Companies shut down, side projects stop being renewed, personal sites lapse when a domain bill goes unpaid. It rarely fails cleanly — a lapsed domain is often picked up by a parked-page operator, so your bookmark now points at an advert rather than a 404.

4. It went behind a login or a paywall

The page exists and the URL resolves, but you cannot read it. News sites tighten their meters. Platforms that once served posts to logged-out visitors stop — X is the obvious case, where links that used to render for anyone now show a sign-in wall.

This one deserves special mention because automated checkers call it healthy. The server returns 200. The content is still unreachable. Any tool that only reads status codes will tell you this bookmark is fine.

5. The platform changed the rules

Whole classes of link die at once when a platform makes a decision. An API closes and every third-party viewer built on it goes dark. A site adds bot detection and archive services can no longer reach it. Nothing about your bookmark changed; the ground underneath it did.

6. The media expired, even though the post did not

The quietest failure of the six. The post is fine, the text is fine, and the image is a grey box. Platforms serve media from CDNs behind signed URLs that expire on a timer, so anything that stored the URL rather than the file now holds a dead reference inside a living page.

What you can actually defend against

Be honest about the split, because any tool promising to solve all six is overpromising:

• Fully defensible — deletion, URL changes, dead sites and expired media, if you kept a copy of the content at the time you saved it rather than a pointer to it.
• Partly defensible — paywalls and logins. A copy taken while you could still read it survives. A copy you never took does not.
• Not defensible — nothing, once you hold the text and the images. Platform policy cannot reach back and take a file off your disk.

Which is the entire argument for copying rather than linking. A bookmark depends on the goodwill and continued existence of a stranger. A copy depends on you.

A proportionate response

You do not need to archive the web. Most links genuinely are disposable — you read the thing, you are done, and a dead bookmark costs you nothing. The ones worth copying are the ones you intend to return to: reference material, research, anything you plan to quote, anything you would be annoyed to lose.

For what is already in your browser, the useful first step is finding out how much of it has died. Our free bookmark checker runs entirely in your browser, uploads nothing, and usually finds more than people expect.

How fast it happens

The research on link rot is consistent and worse than intuition suggests. Studies of web citations repeatedly find that a large minority of links stop resolving within a few years, and that the curve is steepest early — a link is most likely to break in its first couple of years, not its tenth.

Social posts decay faster than ordinary web pages, for reasons specific to them: accounts are deleted and suspended, people make old posts private in bulk after a job change, platforms restrict logged-out access, and moderation removes content retroactively. A five-year-old bookmark folder full of social links is mostly archaeology.

You can measure your own rate rather than trusting an average. The free dead bookmark checker runs in your browser, checks your exported bookmarks and reports which are gone.